Time series momentum: the academic evidence behind trend-following strategies

Time series momentum is a systematic strategy that takes long positions in assets with positive recent returns and short—or zero—positions in assets with negative recent returns. Unlike cross-sectional momentum, which ranks assets against each other, time series momentum treats each asset independently and asks a single question: has this asset trended up or down over the lookback period?

What time series momentum is

The strategy was rigorously documented by Moskowitz, Ooi & Pedersen (2012), Time Series Momentum, Journal of Financial Economics, who examined 58 liquid futures contracts across equities, bonds, commodities, and currencies over 25 years (1985–2009). They found that a 12-month lookback with a 1-month holding period produced positive, statistically significant returns in nearly every asset class—independent of cross-sectional momentum effects.

The core signal is straightforward: if an asset's return over the past 12 months is positive, go long; if negative, go short or stand aside. This binary framing captures the directional component of trends without requiring complex forecasting.

Time series momentum differs from simple trend following primarily in its academic framing and its emphasis on each asset's own history rather than relative ranking. In practice, many systematic strategies blend elements of both approaches. The key insight from Moskowitz et al. is that the autocorrelation in returns is statistically detectable across a remarkably diverse set of markets.

How it works

Implementation follows a consistent pattern across academic studies and practical applications:

1. Lookback period. The prior 12 months is the most commonly studied formation window, excluding the most recent month to avoid microstructure-driven reversals.
2. Position sizing. Moskowitz et al. size positions inversely proportional to each asset's volatility, ensuring that each trade contributes roughly equal risk. This is distinct from equal notional exposure.
3. Universe breadth. Performance improves with a broader, more diverse universe. A strategy spanning equities, fixed income, commodities, and currencies is more robust than a single-asset-class implementation.
4. Rebalancing. Monthly rebalancing is standard, though quarterly rebalancing can reduce transaction costs at modest cost to returns.

The strategy is inherently long–short in its pure academic form. In practice, many self-directed investors implement a long-only version by going to cash—or bonds—on negative-signal assets rather than shorting them.

What the evidence shows

Moskowitz, Ooi & Pedersen (2012) reported a Sharpe ratio of approximately 1.28 for the composite time series momentum strategy across all 58 instruments, from 1985 to 2009. This compares favourably with the 0.38 Sharpe ratio of the buy-and-hold benchmark across the same universe over the same period.

Importantly, the strategy performed well during equity market crises. In 2008, when global equities fell sharply, the time series momentum strategy generated strongly positive returns, driven by short positions in falling equity futures and long positions in rising bond and currency markets. This crisis alpha is a key differentiator versus long-only equity exposure.

Hurst, Ooi & Pedersen (2017), A Century of Evidence on Trend-Following Investing, Journal of Portfolio Management, extended the analysis back to 1880 across 67 markets. They found consistent positive returns across ten-year sub-periods, including the Great Depression, World War II, and the 1970s stagflation era—reinforcing the view that the premium is structural rather than sample-specific.

The correlation between time series momentum and traditional asset classes is low to negative, making it a valuable diversifier within a multi-asset diversification framework.

Limitations and trade-offs

Whipsawing. In range-bound, sideways markets, trend-following strategies generate repeated small losses as positions are entered and reversed without capturing meaningful trends. Prolonged low-volatility, trendless environments are the primary source of underperformance.

Crisis lag. Time series momentum signals are backward-looking. At major turning points—such as the start of a bear market—the strategy may remain long for the first weeks or months of a decline before the signal turns negative.

Crowding and capacity. As trend-following has grown into a multi-trillion-dollar industry, there are legitimate questions about whether historical alphas will persist at scale. Liquid instruments in large, well-traded markets remain less susceptible; smaller or illiquid markets face greater capacity constraints.

Volatility of the strategy itself. While the long-run Sharpe ratio is attractive, shorter periods can produce material losses. Investors should review maximum drawdown figures for any trend-following strategy before committing capital.

Time series momentum in pfolio

pfolio uses time series momentum signals as a core component of its rules-based portfolio construction. Assets are evaluated against their own trailing return history, and allocation weights shift accordingly at each rebalancing. The approach closely follows the academic literature—volatility-scaled, multi-asset, and rebalanced systematically—without discretionary overrides. More detail on the methodology is available at how we build portfolios.

Related articles

• Momentum investing
• Adaptive asset allocation
• Portfolio diversification
• Maximum drawdown
• Sharpe ratio